This is all true but to sit there and say that Ram speed doesn’t matter like this other guy did is just ridiculous.
The main problem I have with misinformation when it comes to computers is people read things they don’t understand and interpret it a different way and then spread that knowledge to others as if it’s fact
The information that shows RAM speed doesn’t matter isn’t bad information, it just is often misinterpreted and not applied correctly.
Remember when GN did that video on AIO orientation on how not to kill your pump? People watched it (or parts of it) and got the wrong takeaway. Instead of figuring out the minutiae of why he was saying what he was saying, and how it applies to actual use case scenarios, they ran with “you’re mounting your AIO wrong!” when people were mounting 360s vertically with pumps/coldplates in their usual locations.
Jayztwocents among others had to make videos just to clarify that viewers didn’t understand it correctly, and then explained that it wasn’t actually wrong, and as long as your pump is lower in any way, it’s fine. And in most cases with 360s, there’s no way to do it differently anyway.
So as usual its people presenting data and people interpreting it wrong, being hostile and aggressive on internet forums parroting something they didn’t understand.
Sweet spot with AMD at least is 3600, the benchmark proves performance drops after that…
Apparently the issue is getting the 1:1 FCLK
3200mt/s == 1600mhz
3600mt/s == 1800mhz
4000mt/s == 2000mhz
Getting FLCK to work at 1600mhz-1800mhz is easy; 2000mhz is hard.
You mean the tomshardware that shows that 3600 RAM is the best benchmark?
It’s disprove what you said… on the site you claim to trust…
Search it for yourself… “RAM Benchmark Hierarchy: Fastest DDR4 Memory Kits for AMD and Intel”
Note what site that is from… now tell me again how it’s ‘misinformation’
Get the FLCK outta here… 
Well I use Intel right now so, done.
You can run them at higher speeds than 4000mhz, but then you decouple it from FCLK so really the limit is 4000mhz since you can’t get FCLK (and lose latency) to past 2000 unless you have some golden chip.
Same old story … always end up at a hardware roadblock. We have come a long way from 50 Mhz Intel P60 … Ah those were the days… when the next chip would double what we have now… and that turbo button… games wouldn’t be playable…
I never had a turbo button, but I remember my cousin had a 486DX2/66 and in some games he had to turn the turbo button on (which set it to 33) because the games went insane at 66mhz.
Be glad FPS isn’t locked to frequency anymore, lol
I’m just glad we don’t have the weird days where you need to go in and change the virtual memory just to get Tie Fighter to run lol.
I had a boot disk for that
Depends on what you want to do. If you want to work on Photoshop or brag in benchmarks then you want to get 4000 mhz+ RAM.
If you want the fastest gaming performance then 3200 mhz is fine but you have to tweak subtimings:
https://www.igorslab.de/wp-content/uploads/2021/03/Cyberpunk-2077-Kabuki-Market-FPS-1920x1080-Raytracing-Ultra.png
As you can see 3200 mhz with proper sub-timings is on par or faster than 4000 mhz with tweaked sub-timings. If you play above 1080p most of time your memory will matter much less.
I personally got a 15% boost tweaking by fixing RAM sub-timings on my 10700 in World of Warcraft.